Warning
π¨ Important Notice: This current repository and the Taxis-Vis-Frontend are put on hold. The goal was to see what is possible to do with today tools on the Javascript end side coupled with Python backend for reproducing Taxis-VIS. Now it touches enough yet is not deleted because could be (re-)used. Cheers! @Simon.
The Data Analysis Backend is a Django + Pandas service that performs analytics on taxi trip data.
Once the Taxis Vis Frontend filters taxi trips, it sends a subset of trips here for statistical and **graphical
** analysis,
including histograms, box plots, scatter plots, and time-series visualizations, to name a few.
Note
The Geospatial backend is no longer needed since DuckDB-WASM handles spatial queries directly in the
frontend.
This backend is strictly for data analysis & visualizationβnot spatial filtering.
- Python (>=3.8)
- Django (installed via
uv
orpip
) - (Recommended) UV for seamless virtual environment management
- Pandas (for handling data operations)
1οΈβ£ Clone this repository:
git clone https://github.com/VIDA-NYU/Taxis-Vis-Data-Backend.git
cd Taxis-Vis-Data-Backend
2οΈβ£ Install dependencies using UV:
uv lock
uv sync
3οΈβ£ Run the Django server:
# With UV (recommended)
uv run python manage.py runserver
# Or manually if using pip/venv (though make sure to be in the correct environment)
python manage.py runserver
π‘ By default, the backend runs on http://127.0.0.1:8000.
1οΈβ£ User applies filters in the Frontend (Taxis Vis UI).
2οΈβ£ Frontend sends a filtered subset of trips (CSV) to this Django backend.
3οΈβ£ Django processes the CSV using Pandas and generates Plotly-compatible JSON for visualization.
4οΈβ£ Frontend receives the JSON and renders the requested charts dynamically.
- Frontend (React) README β The user-facing interface that triggers these analysis requests.
- Original Taxis Vis Paper (IEEE) β Research behind the system.
Happy Analysing!
The Taxis Vis Team π